kern/60963: eqos Ethernet on VisionFive not working well

["bsiegert%NetBSD.org@localhost via gnats" <gnats-admin%NetBSD.org@localhost>]

>Number:         60963
>Category:       kern
>Synopsis:       eqos Ethernet on VisionFive not working well
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Nov 04 19:10:00 +0000 2026
>Originator:     Benny Siegert
>Release:        NetBSD-current (11.99.6)
>Organization:
	The NetBSD Foundation
>Environment:
	Architecture: riscv
	CPU:   rv64imacu
	Model: StarFive VisionFive V2
>Description:

See the thread at https://mail-index.netbsd.org/current-users/2026/10/31/msg048031.html.

I updated my VisionFive (riscv64) to the latest current yesterday, and now the
Ethernet interface has issues. ssh into the machine hangs after


debug1: expecting SSH2_MSG_KEX_ECDH_REPLY

This is after connection has been established. It reproducibly hangs at this
stage.

Meanwhile, on the serial console, I see kernel messages like this:

[ 689.5963642] eqos0: TX [250] tx error: 0x10009000
[ 689.5963642] eqos0: TX [252] tx error: 0x10009000
[ 689.5963642] eqos0: TX [2] tx error: 0x10009000
[ 786.5986961] eqos0: TX [11] tx error: 0x10009000
[ 786.5986961] eqos0: TX [16] tx error: 0x10009000
[ 786.5986961] eqos0: TX [23] tx error: 0x10009000

After the reply from Nick Hudson, I went ahead and bisected the last six months
of kernel changes, narrowing it down to this commit:

0311e19ca67faa96d414b94850a708a2888df7f3 is the first 'bad' commit
commit 0311e19ca67faa96d414b94850a708a2888df7f3 (HEAD)
Author: jmcneill <jmcneill%NetBSD.org@localhost>
Date:   Sat Jun 13 15:38:29 2026 +0000

    eqos: enable COE by default

 sys/dev/ic/dwc_eqos.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

I see no mentions of the new "failed to enable COE" error message, so I suppose
it gets enabled and doesn't work.

 visionfive# dmesg | grep eqos
[     1.000000] eqos0 at simplebus1: DesignWare EQOS ver 0x52 (0x41)
[     1.000000] eqos0: hw features 1a2173f7 09845904 01000000 00000000
[     1.000000] eqos0: Ethernet address 6c:cf:39:00:81:5a
[     1.000000] mcommphy0 at eqos0 phy 0: Motorcomm YT8531 GbE PHY
[     1.000000] eqos0: interrupting on plic0 irq 7
[     1.000000] eqos1 at simplebus1: DesignWare EQOS ver 0x52 (0x41)
[     1.000000] eqos1: hw features 1a2173f7 09845904 01000000 00000000
[     1.000000] eqos1: Ethernet address 6c:cf:39:00:81:5b
[     1.000000] mcommphy1 at eqos1 phy 0: Motorcomm YT8531 GbE PHY
[     1.000000] eqos1: interrupting on plic0 irq 78
[    19.970200] eqos0: TX [83] tx error: 0x10009000
[    21.470234] eqos0: TX [89] tx error: 0x10009000
[    24.470302] eqos0: TX [94] tx error: 0x10009000
[    30.470455] eqos0: TX [98] tx error: 0x10009000
[    42.470760] eqos0: TX [101] tx error: 0x10009000
[    66.471370] eqos0: TX [115] tx error: 0x10009000
[   114.472590] eqos0: TX [136] tx error: 0x10009000


>How-To-Repeat:
	Boot from a recent -current after the above commit. Try to log in over
	ssh.
>Fix:
	I wish I knew :/